dostana We do extensive testing before deploying kernel updates. While I acknowledge that the 5 series of kernels seem to have had poor QA compared to the later 4 series kernels, we can't test for everything and do honestly catch most things before they get synced to Shannon (stable).
We have been unable to bring the 4 series LTS kernel to a newer version because every LTS after 4.9 has caused huge regressions for older hardware. 4.14 broke a lot of intel machines. 4.19 broke older AMD GPUs because of the massive changes to their GPU drivers. 5.4 is the next LTS release and still has had some major regressions since release. My hands are tied. I can't go back to anything newer than 4.19 because they are EOL and that would wreck hardware compatibility for new GPUs, CPUs, and wireless cards. And I can't bring LTS forward because it will break the people we maintain it for.
systemd
is a HUGE exception to our normal updates. It's tied into many different things and there will simply be niche use-cases that users rely on that we aren't testing for. We can't catch everything. And there was nothing we could do to make the update and smoother, since systemd couldn't reexec itself.
While I sympathize, having nuked a few installs by accident in my early years using Linux, we do everything we can to test kernels between releases and we were very thorough testing systemd. We feel that we did everything we could to let users know that the systemd update would require a hard-reboot as well.